reading ai (or hai) ragion for a ragion. This ambiguity is especially common where the same letter would occur twice in the full and unelided form, e.g. tutti may stand for tutti i; Gente avara for Gente è avara (Inf. xv. 68); Quali fioretti for Quali i fioretti (Inf. ii. 127); contra for contra (or -tro) a, &c.

(b) The article is often written in the form El (as in 'I' habitually) so that El may stand either for Il or E il; e.g. Inf. xxv. 66, 74, &c. (See note on the latter passage.)

(c) The constant tendency to fusion or irregular division of words is a source of endless difficulty. We meet for instance with such a form as perokivegio for però ch' io veggio at Inf. xv. 116 in 'C', and though in this particular instance the meaning is clear enough, this is often not the case, e.g. Senzarostarsi may stand for Senz' arrostarsi, or Senza rostarsi in Inf. xv. 39; colle or cholle may be for colle or for ch'ho le in xxx. 81. The following readings in Par. xxxii. 26 must be considered altogether indeterminate on MS. evidence: devoti, de' voti, di voti. The interchange of dio, idio and iddio is perhaps in any case scarcely worth noticing, but certainly not when the previous word ends in i as in Inf. vii. 19, 87, &c. In Purg. iv. 138, some Edd. give ed alla riva, but though e dalla riva is doubtless right, the MSS. would not help us much to decide. Nor again would they do so in reference to the true reading convolto, as contrasted with con volto, in Inf. xxi. 46; or between a dir and ad ir in Purg. xx. 119.

(d) No stress can be laid on MS. evidence where the difference of reading depends on the doubling or not doubling of a consonant. Indifference to this point of orthography is a common feature of the Lombardo-Venetian dialects, and as

+ The same difficulty is met with in MSS. of the New Testament, and has often been overlooked by 'worshippers of the letter,' who sometimes would have us adopt a reading giving little or no sense on the strength of an almost infinitesimal difference of spelling in a MS. three or four centuries later than the autograph. Readings, for example, depending on the mere difference of o and w, and e, &c. are often practically indeterminate. See inter al. Field's Otium Norvicense Pars Tertia, pp. 148–9, "It should be borne in mind that IAE and EIAE are rather different spellings than different readings," and he proceeds to give instances where one is written when the other is certainly intended. If this be so, the distinction cannot be insisted on coûte que coûte elsewhere.

far as my observation extends, a very large number of MSS. exhibit this and other dialectic peculiarities of those provinces. If in conjunction with this we remember that some scribes think that i must be inserted regularly after c and g to make them soft (just as others similarly insert to make them hard), we shall see that the interesting variants in Par. xi. 138, coreggier and corregger become absolutely indeterminate. Even the distinction between the different readings re giovane and re Giovanni in Inf. xxviii. 135 would according to the practice of some scribes become indistinguishable, when we further take into consideration the frequent carelessness as to terminal e and i (presently to be mentioned) and the close resemblance of these two letters in some MSS.

(e) The neglect of the superposed mark of abbreviation (~), indicating the omission of some letter or letters (most commonly of ), and indeed sometimes its meaningless insertion, is another very common source of confusion. Hence inter alia the distinction between the third person singular and plural of verbs is often obliterated, e.g. avea and avean, which latter would be commonly written aveà, but the superposed mark has sometimes been forgotten or has faded. An important case occurs at Purg. xxii. 5. Another common result is the confusion of modo and mondo; some MSS. writing modo when we feel sure they meant modo, i. e. mondo, and others modo where the sense requires modo. Without going so far as to describe this as a regular case of 'indeterminate' reading, we cannot at any rate feel quite sure that a given MS. always intends the actual word which it presents in such cases, for these words are interchanged so often that we cannot but suppose accident to have originally played some part in the matter. A well-known and very interesting case

Other curious irregularities of orthography in respect of these letters are noted at Inf. xxx. 148.

In a MS. in the Taylor Institution Library at Oxford of Cecco d' Ascoli's Acerba, the chapter ominously headed ‘Qui in questo capitolo deride Dante dicendo che non si deve scriver fabule' commences thus :

Qui non si canta al modo delle rane

Qui non si canta al modo del poeta
Che finge imaginando cose vane, &c.

where it is quite clear that modo is in no sense a variant, but a sheer blunder for modo. This confusion is especially common in the Bodleian MS. 'I'.

is that of Inf. v. 102. The following are other instances that I have noticed of this interchange, Inf. iv. 75, x. 64, xi. 61, xviii. 30, xxviii. 21, Purg. xv. 18, Par. ix. 108. The slightly different word moto is interchanged with modo or mondo, in Inf. ii. 60 and Par. xxvii. 106. The interesting variants Dante and Da te in Par. xxvi. 104, are another case in point.

I next add a few cases of letters which are often confused from their close resemblance in some handwritings, and first

(f) final i and e (i's not being dotted). I am not speaking of cases where the interchange merely represents archaic usage, as in the singular person-terminations of the conj. mood of verbs, but of cases like gente, parte, &c., which very often stand for genti, parti, &c., e.g. Inf. xxiv. 144 and xxvi. 90, (where see notes), or when the second person singular and plural of verbs are thus confused. This for instance tends to obscure the evidence in an interesting passage suggested to me in a letter by Dr Witte for examination as a testpassage, viz. Inf. v. 120, where he was strongly of opinion that conoscesti and not conosceste was the true reading, the latter being as he thought an obvious lectio facilior". I found, however, after examining a large number of MSS. that the difference was too slight and uncertain to lay much stress upon it.

A combination of the sources of uncertainty mentioned under (a) (e) and (ƒ) will shew how very slight and unreliable is the difference in MSS. between the various readings in Inf. xxii. 101:

Si che (or ch' ei) non teman (i.e. temã).

Si che (or ch' ei) non tema.

Si ch' i' (or io) non tema.

Compare come, com' e', com' el, com' io, parlava in Purg. viii. 94.

(g) Another common interchange is that of final m and n. Very often it is clear that the substitution of one for the other is accidental, and there is no doubt from the context as to the word intended. But as this tends to disguise the dis

'See on this Scartazzini's note on Purg. xx. 100, and Nannucci, Verbi Ital. p. 284.

We may compare the interchange of sing. and plur. due to a similar consideration in Inf. xxiii. 89 (see note in collations).

tinction between the first and third persons plural of verbs, and as in some cases either person may be possible, no stress can be laid on the evidence of some MSS., at any rate when this interchange is frequent (as e. g. in CDI), to support either reading. A case occurs at Purg. xxii. 5 (already referred to) where we have three possible readings, all of which occur in different editions, avea, aveam and avean. Another important case occurs in Purg. xxi. 19. (See also Inf. xxii. 70, and note on that passage in the collations.)

(h) Again c and t are in many MSS. nearly undistinguishable, and in some cases, we may say the difference between c, e, t, and even i (when it is remembered that i's are not generally dotted in MSS.) becomes almost evanescent. Hence the false reading centri for cencri in Inf. xxiv. 87, which is found in many MSS. and repeated in the Aldine and other Editions. This ambiguity may perhaps have facilitated the interchange of zenit (i.e. cinit or cienit) with tiene in Par. xxix. 4.

(i) One of the commonest and most puzzling sources of 'indeterminateness' in MSS. is the similarity of up and down strokes, hence n and u (or v) are sometimes extremely difficult to distinguish, and this combined with the absence of clear divisions between letters (and even words) and also of the dots to 's, involves other letters and combinations of letters in the same ambiguity, such as m, in, ni, iu, ui, iv and vi (see notes on Inf. xxiii. 78, 143), all of which might be disguised in such a common form as u. When we have (as often) four of such strokes together instead of three-thus — (or even five sometimes), the number of possible combinations that they may represent is much increased. To take a few out of almost numberless instances of this difficulty. We are frequently unable to decide whether a MS. means to read nostro or vostro, e.g. Inf. xxvi. 115, xxviii. 5, xxxiv. 124, Purg. xiv. 126; or again nosco or vosco in 1. 105 of the lastnamed Canto. In this way the words vinto and unito, and even giunto in some MSS. (since some scribes regularly omit the initial g in such words, and would write iunto), may be undistinguishable. Hence I cannot but regard the inter

• See note in the collations at Inf. xxv. 67. This probably accounts for the frequent occurrence of ire and gire as variants: e.g. Inf. xxvi. 141, xxxi. 124, 141, &c.

esting variants in Par. xix. 141 as indeterminate, since auusto may stand for a visto, or aiustò, i.e. aggiustò. So again use may have been taken for vise (visse), iuse, or iùse, i.e. giunse, and by this means readings apparently so different as Visse and Giunse in Purg. xxi. 101 may possibly have arisen from an accidental confusion, since the mark over ise might stand either for an omitted, or for another s. This might also be the case with vinse and giunse in Inf. v. 72, and with avvinse and aggiunse in Inf. xvii. 96,, since by a combination of the different sources of confusion which we have mentioned either might arise from an original aunse. As I remarked about modo and mondo, it is scarcely likely that these interchanges would be so common unless they had been aided by, or had arisen from, the accidental resemblance. The curious and widely spread false reading lieve in Inf. xxxiii. 26 may thus easily have arisen accidentally from lune.

(j) Lastly, I will mention a few pairs of often-recurring words which, from their similarity no doubt, are very often confused or interchanged. Such are the following :—alto and altro (very commonly in 'B'), e.g. Inf. i. 9, &c. (see note on Inf. xxiv. 135), and similarly mente and mentre (especially in 'C'); also effetti and affetti in several places. Note especially the interesting passage in Purg. xi. 3, where the disputed reading is recognized by the author of the Ott. Comm. as early as 1333. So also effettuoso and affettuoso in Inf. v. 87. (See note on this passage, where numerous instances of this confusion are collected.) Again, cagion and ragion are almost habitually confused, this interchange being facilitated by the similarity of sense as well as of form (e.g. Purg. xiii. 20, xviii. 65, xxii. 30, xxxi. 90, Par. vii. 101, xi. 21, &c.). Another pair of words constantly confused is caro (or charo) and chiaro and their derivatives, as in the disputed line Par. xxv. 33. Add also Inf. xxxi. 28, Purg. i. 75, and Par. xxii. 146 (though of course here chiaro must be right, and caro is a mere blunder).

The various ambiguities and uncertainties here enumerated have obliged me to abandon several interesting testpassages as 'indeterminate,' on which I had collected a large number of references to MSS., and concerning which in many cases the evidence of MSS. would have been peculiarly welcome, if it could have been given with no "uncertain sound."